KNN_Timeseries_Assignment(1).docx

Week1 Assignment KNN Time Series Mohit Manjaria 9/24/2020 ALY 6020 Predictive Analytics Week1 Assignment KNN Time Series Mohit Manjaria Instructor: Marco Montes de Oca Winter 2021 January 28th 2021 Northeastern University Introduction In this assignment, a dataset of search interest of all categories for Predictive analytics term from January 2061 to January 2021 is selected for K-Nearest Neighbors Analysis. In this dataset, the numbers represent search interest relative to the highest point on the chart for the given region and time. A value of 100 is the peak popularity for the term. A value of 50 means that the term is half as popular. A score of 0 means there was not enough data for this term. Analysis In this part, time series forecasting with KNN regression will be performed. According to the auto-regressive model and model requirements that are given, three dimension, four dimension, five dimension and six dimension models with K values from 1 to 10 will be explored. Step 1: Installing libraries library (tinytex) library (neighbr) library (readr)

library (tsfknn) library (zoo) ## ## Attaching package: 'zoo' ## The following objects are masked from 'package:base': ## ## as.Date, as.Date.numeric library (ggplot2) Step 2: Importing the data and checking the data knnt <- read.csv ( "C:/Users/mohit/Downloads/multiTimeline.csv" ) head (knnt) ## Week Predictive.analytics...United.States. ## 1 1/31/2016 60 ## 2 2/7/2016 54 ## 3 2/14/2016 58 ## 4 2/21/2016 53 ## 5 2/28/2016 43 ## 6 3/6/2016 54 dim (knnt) ## [1] 259 2 summary (knnt) ## Week Predictive.analytics...United.States. ## Length:259 Min. : 9.00 ## Class :character 1st Qu.: 42.00 ## Mode :character Median : 52.00 ## Mean : 53.07 ## 3rd Qu.: 63.00 ## Max. :100.00 #KNN Model pred <- knn_forecasting (knnt $ Predictive.analytics...United.States., h= 1 , lags = 1 : 2 , k= 1 ) pred $ prediction ## Time Series: ## Start = 260 ## End = 260

## Frequency = 1 ## [1] 34 pred $ neighbors ## [1] 259 #Plotting time series autoplot (pred, h= 1 ) #Calculating accuracy for k = 1 ro <- rolling_origin (pred, h= 1 ) ro $ global_accu #(Evaluating Using RMSE, MAE, MAPE) ## RMSE MAE MAPE ## 11.00000 11.00000 32.35294 ro $ predictions ## h=1 ## [1,] 45 ro $ h_accu ## h=1 ## RMSE 11.00000 ## MAE 11.00000 ## MAPE 32.35294

#Calculating Euclidean Distance install.packages(“philentropy”) library(philentropy) #knn.dist(knnt, dist.meth = “euclidean”, p = 2) #For n=2 or K = 3 pred <- knn_forecasting (knnt $ Predictive.analytics...United.States., h= 1 , lags = 1 : 2 , k= 3 ) pred $ prediction ## Time Series: ## Start = 260 ## End = 260 ## Frequency = 1 ## [1] 44.66667 pred $ neighbors ## [1] 259 218 206 #Plotting time series autoplot (pred, h= 1 ) #Calculating accuracy ro <- rolling_origin (pred, h= 1 ) ro $ global_accu #(Evaluating Using RMSE, MAE, MAPE)

## RMSE MAE MAPE ## 16.33333 16.33333 48.03922 ro $ predictions ## h=1 ## [1,] 50.33333 ro $ h_accu ## h=1 ## RMSE 16.33333 ## MAE 16.33333 ## MAPE 48.03922 #For n=2, k = 5 pred <- knn_forecasting (knnt $ Predictive.analytics...United.States., h= 1 , lags = 1 : 2 , k= 5 ) pred $ prediction ## Time Series: ## Start = 260 ## End = 260 ## Frequency = 1 ## [1] 50.6 pred $ neighbors ## [1] 259 218 206 149 256 #Plotting time series autoplot (pred, h= 1 )

#Calculating accuracy ro <- rolling_origin (pred, h= 1 ) ro $ global_accu #(Evaluating Using RMSE, MAE, MAPE) ## RMSE MAE MAPE ## 20.00000 20.00000 58.82353 ro $ predictions ## h=1 ## [1,] 54 ro $ h_accu ## h=1 ## RMSE 20.00000 ## MAE 20.00000 ## MAPE 58.82353 #for n = 2, k = 7 pred <- knn_forecasting (knnt $ Predictive.analytics...United.States., h= 1 , lags = 1 : 2 , k= 7 ) pred $ prediction ## Time Series: ## Start = 260

Your preview ends here

Eager to read complete document? Join bartleby learn and gain access to the full version

Access to all documents
Unlimited textbook solutions
24/7 expert homework help